Research On PID Controller Optimization Based On The Data-Driven Technique

With the development of modern industry, the process of production is more and more large and complex, the operating point varies widely, which makes it dificult to build models. Data-driven as a technique that doesn’t require the object model, and only requires I/O data to complete controller design has become a hot research. In this paper,we use the I/O data of the system directly to optimize the PID parameters online for nonlinear system. Firsly, we use K-NN(K-Nearest Neighor) algorithm to elect information from the system database that matches the current system status, and then using the linear weighted average method to obtain the PID parameters. On the basis, according to the optimization criterion function, it achieves the PID parameters adjustment online and gives out the relevant database update principle. Through several nonlinear system simulation experiments, it conformes the effectiveness of the optimization algorithm.Secondly, the proposed method is applied to the superheated steam temperature control system.Respectively using Matlab/Signal Constraint toolbox and the proposed method to optimize paramters offline or online for the superheated steam temperature system controller under different load. And then we get the optimized values obtained by the two methods as the PID controller parameters and conduct step disturbance test, the results cofirm the effectiveness of the proposed algorithm further. Then we take a robustness test for 75% load of the system,and the result shows that it has certain robustness. Finally, the optimization algorithm is proposed by Matlab/GUI graphical user interface.User can operate it even if he dosen’t understand the algorithm and observe the out curve of the system and the corresponding value of the parameter optimization in both the traditional PID control and PID optimization based on data-driven control.